Constructions and Frames - Volume 10, Issue 2, 2018

This article studies the pseudo-coordination [gå ‘go/walk’ och ‘and’ V]. The construction has several meanings and it also has subordination counterparts in Modern Swedish, unlike most Swedish pseudo-coordinations. Our diachronic study shows that [gå och V] cannot readily be reduced to the verbs in isolation and that synchronic lexicocentric perspectives based on syntactic (re)configurations cannot capture the constructional meaning such as the assumed inference of ‘surprise’ or ’unexpectedness’. We argue that a detailed analysis of the historical development makes the picture clearer.

In the development of [gå och V], item-based analogy continuously facilitates new verbs in the V slot. At a certain stage, there is a mismatch between the agentivity of the construction and the non-agentivity of events denoted by the second verb. This mismatch is resolved by the override principle that forces non-agentive verbs to be interpreted agentively and promote a more abstract and lexicalized version of the construction. The exemplar-based view to constructions proposed by Bybee (2010, 2013) seems favorable, since frequent exemplars of [gå och V] allow for redundant or marginal features to serve as the model for novel expansions of the construction.

The paper deals with the French morphological prefixation pattern [non-N] (non-qualification ‘non-qualification’, non-Italien ‘non-Italian’, and non-ville ‘non-city’). It discusses the form/meaning asymmetry displayed by this pattern and its compositionality. It is shown that the general pattern [non-N] actually corresponds to three distinct subconstructions, i.e. distinct form/meaning pairings. Although pragmatic factors may be seen as presenting a challenge to the compositionality of these constructions, it is argued that [non-N]s must be seen as compositional as long as compositionality is defined not only in terms of truth-conditional semantics, but also of pragmatics.

This paper offers a fresh perspective on (restrictions on) aspectual coercion, thereby focusing on the essentially epistemic import of aspectual constructions. The case study that I will discuss is the unexpected use of the simple tenses for ongoing event reports in sentences involving full-verb inversion. I will argue that this attestation of the simple present/past in inverted sentences can be analyzed as a kind of aspectual mismatch between the higher-order construction and the embedded tenses. Yet at a more basic, epistemic level of analysis, there is no mismatch: the full-verb inversion construction and the embedded tenses are similar in the sense that both report events that are conceived of as fully and instantly identifiable.

Prior studies suggest that language users perform motoric simulations when construing action sentences and that verbs and constructions each contribute to simulation-based representation (Glenberg & Kaschak 2002; Richardson et al. 2003; Bergen et al. 2007; Bergen & Wheeler 2010). This raises the possibility that motorically grounded verb and construction meanings can interact during sentence understanding. In this experiment, we use the action-sentence compatibility effect methodology to investigate how a verb’s lexical-class membership, constructional context, and constructional bias modulate motor simulation effects. Stimuli represent two classes of transfer verbs and two constructions that encode transfer events, Ditransitive and Oblique Goal (Goldberg 1995). Findings reveal two kinds of verb-construction interactions. First, verbs in their preferred construction generate stronger simulation effects overall than those in their dispreferred construction. Second, verbs that entail change of possession generate strong motor-simulation effects irrespective of constructional context, while those entailing causation of motion exert such effects only when enriched up to change-of-possession verbs in the semantically mismatched Ditransitive context. We conclude that simulation effects are not isolable to either verbs or constructions but instead arise from the interplay of verb and construction meaning.

In every-day language use, two or more structurally unrelated constructions may occasionally give rise to strings that look very similar on the surface. As a result of this superficial resemblance, a subset of instances of one of these constructions may deviate in the probabilistic preference for either of several possible formal variants. This effect is called ‘constructional contamination’, and was introduced in Pijpops & Van de Velde (2016). Constructional contamination bears testimony to the hypothesis that language users do not always execute a full parse of the utterances they interpret, but instead often rely on ‘shallow parsing’ and the storage of large, unanalyzed chunks of language in memory, as proposed in Ferreira, Bailey, & Ferraro (2002), Ferreira & Patson (2007), and Dąbrowska (2014).

Pijpops & Van de Velde (2016) investigated a single case study in depth, namely the Dutch partitive genitive. This case study is reviewed, and three new case studies are added, namely the competition between long and bare infinitives, word order variation in verbal clusters, and preterite formation. We find evidence of constructional contamination in all case studies, albeit in varying degrees. This indicates that constructional contamination is not a particularity of the Dutch partitive genitive but appears to be more wide-spread, affecting both morphology and syntax. Furthermore, we distinguish between two forms of constructional contamination, viz. first degree and second degree contamination, with first degree contamination producing greater effects than second degree contamination.